Comparison of Evidence-Based and Traditional Employment Services for Persons with Serious Mental Illness:Implications forSomatic and MentalHealth Care

Continuity and Coordination

Steinwachs DM, Salkever D, Stuart EA, Abrams M, Skinner EA, Wu AW, Salzberg C,

Gibbons B, Mojtabai R, Daumit G

ABSTRACT

Objective: To evaluate the effect of initiation of evidence-based supported employment services (Individual Placement and Support-IPS) on continuity and coordination of mental and somatic health care for persons with serious mental illnesses enrolled in Medicaid and qualifying for state or federal disability.

Data Sources: Maryland Medicaid data were matched to data from the Division of Rehabilitative Services and the Maryland Public Mental Health System to identify the study groups and service initiation dates. Maryland Medicaid data was used to measure continuity and coordination for two years prior to initiation and three years post. Initiation of employment services occurred between September 1, 2006, and December 31, 2007.

Study Design: An intent-to-treat design is used to compare three groups of Medicaid persons with SMI: those initiating supported employment that followed the Individual Placement and Support model (IPS) that met evidence-based fidelity criteria (IPS-F); those initiating supported employment services that may or may not follow the IPS model and were not certified as meeting the IPS fidelity criteria (SE-NF); and those initiating traditional vocational services, emphasizing training that expected to lead to competitive employment (TVS). Using propensity score weighting the three groups are compared on outcomes over a 3 year period including: (a) continuity of care for medications and provider continuity for mental health and somatic care, and (b) coordination of care outcomes within 30 days after hospital discharge, based on hospital readmissions, emergency department visits, and follow-up visits with community doctors.

Data Extraction Methods: Data sets were provided by the Maryland Medicaid Program, the Maryland Division of Rehabilitative Services and the Maryland Public Mental Health System.

Principal Findings: Patterns of significant findings are mixed. Statistically significant positive associations of IPS-F with medication continuity were present in one of three years for antidepressant medication, in two years for antihypertensive medication, and were not present for antipsychotic medications. IPS-F was positively associated with mental health provider continuity but not statistically significant levels. IIPS-F was not significantly associated with somatic provider continuity and all coefficients were negative suggesting less provider continuity. The hospital post-discharge indicators of care coordination were mixed. No significant associations of IPS-F with ED visits within 30 days. There was borderline significance for IPS-F and changes in hospital readmissions years 2 and 3, with lower readmissions followed by higher readmissions. There were no significant associations of IPS-F with at least one physician visit with 30 days after hospital discharge.

Conclusions: Associations between supported employment, specifically IPS-F, as compared to traditional vocational services were neutral or positive for medication continuity and not significant for provider continuity. Associations between supported employment and coordination of care were mixed and not statistically significant. Smaller than expected sample size may have contributed to less statistical significance as might have variations in employment program selection of clients and employment services provided.

Key Words: Supported employment, continuity/coordination of care, chronic illnesses

Acknowledgement: This research was supported by Contract No. HHSA290201000009I from the Agency for Healthcare Research and Quality, US Department of Health and Human Services, as part of the Developing Evidence to Inform Decisions about Effectiveness (DEcIDE) Program.

INTRODUCTION

Nationally, persons with serious mental illnesses (SMI) with serious functional impairment are estimated to include 4.1% of U.S. adults (SAMHSA, 2013). Estimates are that over 80% of persons in the community with schizophrenia are not employed (Salkever, et al., 2007)despite research showing that Individual Placement and Support (IPS), an evidence-based supported employment intervention, can be successful in achieving competitive employment positions for half or more of persons with SMI desiring work (Bond, Drake, and Becker, 2008; Bond and Drake 2014). Traditional employment services in Maryland differ from supported employment and emphasizetraining and non-competitive job experiences (e.g. enclave employment) prior to or instead of placement in a competitive employment environment.

In 2002, the Maryland Mental Hygiene Administration undertook a statewide initiative to provide evidence-based employment services for all persons with SMI wanting to work, largely replacing traditional employment training and placement programs. For persons with SMI, competitive employment represents a major positive step toward recovery and has been endorsed by the President’s New Freedom Commission on Mental Health (2003), the Surgeon General (1999), and the National Institute of Mental Health (1999). We hypothesized that the IPS employment intervention would have positive effects on the individuals’ lives including greater engagement with healthand mental health care as evidenced by improved continuity of care, improved care coordination and reductions inhospital post-discharge readmissions and use of emergency departments (ED). This study compares the experience of persons with SMI receiving supportedemployment services which emphasize immediate job placement,to those initiating traditional employment services that emphasize training prior to job placement.

Background

Individual Placement and Support (IPS) is a version of supported employment that uses a “place and train” approach (Wehman and Moon, 1988) in contrast to traditional practices in which extensive pre-vocational training is provided prior to job placement (“train and place” model). The supported employment model was adapted to meet the needs of persons with SMI and multiple randomized controlled trials have established its efficacy (Bond, Drake and Becker, 2008). In nine RCTs conducted in eight states (including Maryland), IPS employment rates ranged from 27% to >75%, compared to controls that achieved 7% - 40% employment. IPS has eightcharacteristics: (1) involvement in competitive employment from the outset instead of placement in a sheltered or non-competitive work setting,(2) all consumers(the person with the SMI) desiring to work are eligible for IPS, (3) rapid job search (placement frequently within a month), (4) integration of mental health and employment services, (5) attention to consumer preference in the job search, (6) time-unlimited individualized job supports, (7) personalized benefits counseling, and (8) systematic job development (Dartmouth Psychiatric Research Center 2011) .

IPS has been rigorously evaluated over the past two decades. Arecent Cochrane Review (Kinoshita, 2013) identified 14 randomized controlled trials (RCT) in the U.S. and elsewhere that compared IPS with other programs that largely emphasized training for work prior to placement. In all of the RCTs, the primary outcomes were employment, generally assessing how soon competitive paid jobs were obtained, and duration of employment. IPS programs consistently outperformed traditional employment services (Kinoshita, 2013).

A potential concern in the implementation of evidence-based interventions is fidelity to the original evidence-based model. A 1997 study comparedstaffing, organizational features and service characteristics of 26operating IPS programs with the original IPS model as evaluated in RCTs(Bond, et al., 1997). Overall, the study found that significant deviations from the original model were strongly associated with poorer employment outcomes. These findings led to fidelity criteria with a cut-off level defining fidelity and non-fidelityof IPS programs. Since 2006 in Maryland, IPS programs meeting fidelity criteria have received higher reimbursement for some specified supported employment services.

To have an impact on health, IPS would need to improvethe interaction of individuals with SMI and their health care providers. This may be achieved through IPS program design in which employment services providers meet with mental health providers to coordinate services, and throughthe effects of competitive employment on the person with SMI. Recently, Bond and Drake (2014) reviewed research findings on the benefits of competitive employment and found that, “People who obtain competitive employment through IPSenhance their income, self-esteem, quality of life, social inclusion, and control of symptoms.” Their findings suggest that IPS leading to competitive employment is contributing to patient activation, that is, the patient’s willingness and ability to manage their health and health care (Hibbard 2007).

Evidence from prior research has reported IPS effects on mental health service use. Gold, et al. (2006) reported more mental health service contacts for participants in a program that combined IPS and assertive community mental health treatment, compared to those in traditional vocational programs. Another study found fewer hospital admissions among IPS program users than users of traditional employment programs and less time spent in hospitals overall (Burns, et al., 2007). Drake, et al. (2013), and Salkever et al. (2013), also found fewer hospital days, hospital stays, and ED visits for mental health problems, and psychiatric crisis services for the treatment group, in the aforementioned SSDI randomized trial vs. the control group. Activation may be a pathway linking vocational interventions and patient-centered health care outcomes, as suggested by the review of IPS intervention studies result in enhanced income, self-esteem, quality of life, social inclusion, and control of symptoms (Bond and Drake 2014). In this study, we are looking for changes in patterns of care for both mental health and somatic conditions. It is notable that the premature mortality for persons with SMI and disabilities.

The paper presents comparisons among consumers meeting SMI and continuous Medicaid enrollment criteria who entered either an IPS program meeting established fidelity standards (IPS-F), a supported employment program without fidelity (SE-NF) or a traditional employment services program (TVS). We hypothesized that IPS-F would be associated with improved coordination of services as compared to TVS and possibly similar to SE-NF employment program outcomes. IPS-F is expected to:

  1. Increasecontinuity of medications for selected chronic mental and somatic conditions as compared to TVS,
  2. Increase mental health and somatic provider continuity,
  3. Reduce hospital readmissions for all causes within 30 days of discharge,
  4. Reduce emergency department visits within 30 days post-discharge, and
  5. Increase continuity of physician post-discharge care within 30 days.

METHODS

Data Sources

Three data sources were used to identify study subjects andclassify each subject by type of employment services received and date initiated. Maryland Medicaid administrative data were used to identify persons with serious mental illnesses (SMI) by diagnosis who met age, disability status and other enrollment eligibility criteria. The Maryland Division of Rehabilitation Services (DORS) pays for employment services, including supported employment and traditional vocational services. The Maryland Public Mental HealthSystem (PMHS) has paid only for supported employment servicessince the 1990’s but has restricted support to programs following the IPS model (with or without meeting fidelity criteria) since 2009. Supported employment services may be initially paid for by either DORS or PMHS. To identify the study population, files from DORS, PMHS and Medicaid were merged and matched using social security numbers.

The measurement of dependent and independent variables relied primarily on Medicaid enrollment and claims data over a period of two years prior to initiation of employment services and three years post-initiation. All the dependent variables were constructed for 12-month time periods centered on the date of initiation of employment services. The study was approved by the IRBs of the Johns Hopkins Bloomberg School of Public Health, the University of Maryland Baltimore County, and the Maryland Department of Health and Mental Hygiene.

Study Population Selection

The sample selection criteria identified persons with serious mental illness(SMI) diagnoses and state or federal certified disability as identified in the Medicaid enrollment files with at least 10 months per year of Medicaid enrollment over a 4-5 year period, 2 years prior and 3 years post initiation of an employment services intervention. Persons were eligible for inclusion if their first employment services contact occurred between September 1, 2006, and December 31, 2007, which marked thebeginning of statewide implementation of an Outcomes Measurement System (OMS) that is used by the PMHS to track the progress of the patients that they serve. Persons were included in the study if they met all the following selection criteria. The figures in parentheses show the numbers remaining after completion of the step.

  1. Had a serious mental illness diagnosis in calendar years 2006 or 2007 (n=131,820)
  2. Age 20-63 years between 9/1/2006 and 12/31/2007) (n=79,759)
  3. Eligible for disability benefits according to state or federal criteria during the period 7/1/2006 to 12/31/2007 (n=35,253)
  4. Were not Medicare enrollees at initiation of employment services (n=22,556), given that Medicare data were not available for this investigation
  5. Had at least 10 months of Medicaid enrollment in each 12 month period two years before and two years after the employment services initiation period, 9/1/2006 and 12/31/2007 (n=16,321)
  6. Had a matching DORS or PMHS record indicating initiation of employment services between 9/1/2006 and 12/31/2007 (n=618)
  7. Did not receive employment services in the year before the initiation period and lived in Maryland during the same period (n=433)

Serious mental illness was defined as the presence of at least one of the following International Classification of Diseases Version 9 (ICD-9) diagnostic codes, as recorded in the Medicaid data:

295.xxSchizophrenic disorders

296.xxEpisodic mood disorders (mania, bipolar, major depression)

297.xxDelusional Disorders

298.xxOther Nonorganic Psychoses

299.8x, 299.9xOther pervasive developmental disorders

300.xxAnxiety states

301.xxPersonality disorders

302.xxSexual and gender identity disorders (excluding 302.7x, psychosexual dysfunction)

310.xxSpecific nonpsychotic mental disorders due to brain damage

311.xxDepressive disorders, not elsewhere classified

Classification of subjects in three treatment groups

Individuals in our sample of 433 were classified into one of three mutually exclusive groups based on the type of service they received at the first employment service received between 9/1/2006 and 12/31/2007:

  • Individual Placement and Support (IPS-F) programs certified as meeting the criteria for fidelity to the IPS model at time of service initiation (N=136);
  • Supported employment services that place clients in competitive employment (SE-NF); these programs may or may not follow the IPS model and were not certified as meeting the IPS fidelity criteria at time of service initiation (N=171);and
  • Traditional vocational services (TVS) which offer a “train and place” approach (N=126).

Measures

Three categories of care coordination and continuity measures are examined. Continuity of care measures include medication continuity (filling prescriptions regularlyover time), provider continuity (seeing the same provider over time), and visit continuity (seeing a provider at least monthly). Indicators of successful care coordination include quality indicators for hospital post-discharge care, including hospital readmissions within 30 days, emergency department (ED) visits within 30 days, and having a physician visit within 30 days.

Continuity of Care

Medication continuitymeasures the extent to which prescriptions have been filled over the entire year in each drug class, antipsychotic, antidepressants, antihypertensive, and blood sugar regulator medications. Findings for blood sugar regulators are not included due to small numbers of persons taking these medications. The hypothesis is that medication continuity is expected to increase significantly for those receiving IPS-F employment interventions as compared to TVS employment services. Higher levels of antipsychotic medication continuity as compared to low levels have been shown to reduce hospitalizations substantially among persons with schizophrenia (dosReis 2008). In general, persons with treatable chronic conditions experience better outcomes by consistently filling and taking prescribed medications.

Continuity is measured on a zero-to-one scale based on filled prescriptions in each of four drug classes during each year.Using prescription fill dates and days of supply, total days supply in each class is calculated for each year (capped at 365), thendivided by 365 to measure the proportion of days in the year covered by filled prescriptions (Choudhry, et al., 2009). Study subjects mayhave overlapping prescriptionsor multiple prescriptions in a drug class with the result that total days supply could exceed 365 days per year. Overlapping prescriptions might result from medication switching (where there are leftover pills) or being prescribed two drugs to be taken concurrently in one drug class. To identify drugs by class, national drug code (NDC) lists were generated using lists from the U.S. Food and Drug Administration as well as from Maryland’s state Medicaid files which use NDCs as well as American Hospital Formulary System labels ( Prescriptions generally were written for a one month supply.

Provider continuity: In each time period, the provider to which the individual madethe most visits was identified as the principalprovider. The continuity measure is the fraction of ambulatory visits seen by the principal provider (excluding ED visits) out of total ambulatory visits to all providers. Provider continuity was calculated separately for mental health visits and for somatic visits during each year. This measure of provider continuity been used in previous research for Medicaid populations (Stuart 1993) using administrative data and among persons with SMI disorders (Chien 2000). Among persons with mental illness, provider continuity is related to lower hospitalization rates and lower Medicaid costs (Chien 2000) and to reduced risk of mortality (Hoertel2014).

Coordination of care

Care coordination is receiving greater attention with the current emphasis on reducing 30 day re-hospitalization rates under the Affordable Care Act (Kocher 2011). This was not a national priority during the study time period and coordination of post-discharge care was largely a patient/family responsibility to ensure needed post-discharge care was received. Coordination of care variables were constructed to examine post-discharged hospital care, including: seeing a physician within 30 days post-discharge, readmission to a hospital within 30 days post-discharge, and having an ED visit within 30 days post-discharge. Hospitalizations include mental health, substance abuse, and somatic conditions.